Space Missions Engineering Laboratory
Heuristic Methods
Heuristic optimization is a new approach for solving complex problems that overcomes many shortcomings of traditional optimization techniques. Heuristic optimization techniques are general purpose methods that are very flexible and can be applied to many types of objective functions and constraints. The power of Heuristic Methods stays in that no mathematical properties must be satisfied by the functions that model the problem to be solved: mixed search domain can be dealt with, and heuristics lead the search strategies, different depending on the applied algorithm. Many of them can be retrieved from the Evolutionary Computation area: Evolutionary Algorithms and Genetic Algorithms mimic successful strategies found in nature copying the principles on which species develop superior qualities over generations; Simulated Annealing leans on how crystals emerge when materials are cooling and particles "find" a structure that minimizes the energy balance; Particle Swarm Optimization simulates the swarms behaviour: swarm members take advantage from good detected areas from any of the flies in the team to tune each member velocity toward the target, represented by the function optimization; Ant Colony Optimization copies the ants behaviours in following pheromone trails whenever the path-to-food has been detected by a member of the colony; they reveal to be successful whenever applied to combinatorial problems such as the optimal path into graphs detection. It should be remembered that a decision making can be formalised as a graph.
Heuristic methods generally perform global optimization thanks to the population-based feature: the solutions visited at each step are spread all over the search space; therefore local basins are easier avoided.
Our group successfully applied Heuristic Methods to optimization problem solving since many years: non-linear, multivariate and multi-modal problems, such as interplanetary trajectory optimal design, as well as system design optimization with different operative mode have been solved. Those techniques turned out to be well-suited for multi-objective optimization too: not only a large amount of information can be analyzed simultaneously but also quantities of very different nature can be treated together wit no risk of incurring in badly conditioned problem formalisation.
Although population-based techniques are definitely suggested to preliminary prune the solution space and focus future analysis on a very specific zone of the search space, whenever complex scenarios are dealt with a disturbed architecture for the optimization loop is mandatory: Game Theory protocols linked to heuristic methods gave interesting results according to multidisciplinary optimization problems – such as Entry Descent and Landing guidance profiles and configuration design, multi-module space system preliminary design, space mission design activities scheduling.
Powered by CMSimple